Florian Jetter

FYI I'm currently working on removing the Sequence/Dict classes. I suspect there could be a perf gain. I'll report back once it is stable. Either way, we can move forward...

I have a follow up ready that addresses all of the above review comments and removes the sequence/dict tasks. In favor of having a simple review, I'll move forward with...

+1 Let's get a warning out asap. When we'll drop the actual implementation is something we can discuss further but the decision has been made a long time ago to...

@dask/gpu gpuCI appears to be broken again. One example https://github.com/dask/dask/pull/11354 but there are other failures and it looks quite intermittent. Looking at Jenkins this almost feels like a gpuCI internal...

How long will it take to replace the Jenkins infra? I currently feel gpuCI is not delivering a lot of value and is just noise. Would you mind if we...

I'm not very deep into this but from a high level it makes sense for us to reuse the pandas `pd.api.extension.no_default` for all APIs that pass this to pandas directly....

xref There is also an actual functional problem with this besides mere UX https://github.com/dask/dask/issues/9888

> I'll note that with these ideas, the client becomes an increasingly-integral part of the system, since the cluster no longer holds the full graph, nor all of the results,...

We removed a lot of special treatmenents in `check_thread_leak` function to track down leakages. One of the changes was that we now "warm up" the BaseTCPConnector Threadpool, see https://github.com/dask/distributed/blob/2fffe74dde40238a3387d12740ff9a54c16acb9f/distributed/utils_test.py#L1573-L1576 https://github.com/dask/distributed/blob/2fffe74dde40238a3387d12740ff9a54c16acb9f/distributed/comm/tcp.py#L411-L419...

> My understanding is that there are two parts to the annotations problem: > > We need a way to attach annotations to specific expressions > We need a way...